[437] in NetBSD-Development
New kernels
daemon@ATHENA.MIT.EDU (ghudson@MIT.EDU)
Thu Jan 19 13:06:48 1995
From: ghudson@MIT.EDU
Date: Thu, 19 Jan 1995 13:06:24 -0500
To: netbsd-dev@MIT.EDU
I've built new Athena kernels and placed them in
/afs/sipb/project/netbsd-dev/athena-install/kernels. They have the
following difference from the old kernels:
* They use the exec patch so that they can run NetBSD-current
binaries. (The patch is just a diff of granola's machdep.c
and locore.c vs. the 1.0 sources.)
* They add the following options and pseudo-devices:
options SYSVMSG
options SYSVSEM
options SVSVSHM
options KTRACE
pseudo-device bpfilter 4
pseudo-device ppp
I looked over the NetBSD-current autoprobing strategies for AHA and BT
SCSI drivers, and decided that I do not feel qualified to update the
NetBSD 1.0 sources so that a single kernel can handle both drivers.
This means we have a few options:
1. We can continue to have at least two kernels (four until we
can resolve the 3C509 problems) until the next release of
NetBSD.
2. We can snapshot a -current kernel source tree, watch
source-changes for bugfixes, test the kernel for a month or
so, and use that after we're confident that it's stable.
3. Someone else can update the NetBSD 1.0 sources so that a
single kernel will work for both AHA and BT SCSI cards.
I don't like (1) since NetBSD releases come about two years apart, and
it means I have to do twice as much work when making any kernel
configuration changes or rebuilding kernels. It also complicates the
installation process by necessitating that the install script pick a
kernel (using some heuristics which are not necessarily 100% reliable)
rather than simply including the kernel in the distribution.
Unless someone else wants to implement (3), I will probably go with
(2) sometime next term.
